Finite element analysis and artificial neural network for stress distribution of an aircraft model in a wind tunnel

Abstract

Wind tunnels are instrumental in the aerodynamic analysis of aircraft model structures, enabling the replication of real circumstances for better design and performance evaluation. This paper presents a novel enhancement to stress distribution predictions in wind tunnel simulations by combining Finite Element Analysis (FEA) and Artificial Neural Networks (ANN). First, the research focuses on analyzing ANSYS Fluent data, which provides insights into the complex fluid dynamics inside the wind tunnel. The proposed approach combines the best available FEA and ANN techniques regarding prediction accuracy and computational efficiency. Such findings are those that evidence that predictions of real stress levels using ANN are quite near, with RMSE 12%, and, hence, quite accurate. The results indicated agreement between the functions generated by ANN and real stress levels and, therefore, were considered to manifest a very low error percentage. The methodology shows that it is significant for being computationally efficient since the ANN works much quicker compared to the conventional FEA approach. In addition, the methodology is significant in computations since the ANN works quicker than conventional FEA. These results thus indicate that the integrated FEA-ANN approach is beneficial and holds much promise in accurately and efficiently predicting stress distributions. Herewith, the provided method advances engineering simulations by making exact predictions of stress distributions necessary to improve design and structural analysis.